UNIX for Dummies Questions & Answers

When you run a ps -ef and if the status is a Z (zombie) does that mean the same as not responding? (Like a windows machine).

Also has anyone here heard of the program called 'top' (I've found it on our Solaris 7 machines) If you have you might be able to help me. I need to know if there is a way to test the status of a machine or program from the top command.


Might help if I explian what I'm trying to do.

I need to write a script/program to detect when a program isn't running and then restart that program. BUT sometimes it's not the program that is responding it can sometimes be the network port and just gone in a zombie mode and it's detected and all but wont receive info or send any through that port. I need to be able to detect that as well and reset that port if that is the problem.

If anyone has any idea's how this can be done or help with the above. Thanks so much if you can help

A zombie is a program that has completely exited. It is waiting for it's parent process to issue a wait() system call. When you see a zombie in "ps", look at the ppid field. That is the pid of the parent. If you kill the parent, the zombies will become inherited by init which will notice and issue that wait() call. This makes the zombies go away. You really should track down the author of that parent process. This is a bug and it should be fixed. Programs that launch child processes need to be ready for the death of a child.

My knowledge of windows is very weak, but I think a "not responding" program is more like a unix program that is ignoring as many signals as possible. If you do a "kill $pid" on it, it won't go away. You need to do a "kill -9 $pid" on it to get rid of it.

A port cannot be a zombie, so I don't know what you have in mind there.

If the child is not responding to the parent you can issue the -18 SIGCHILD to tell the parent to go out and kill all its children

Of course if you don't have the PID of the parent you can't do this. If the child has been handed off to init and the PPID is (1) then you can't get rid of them until your next reboot.

a zombie process can be identifed when u do a ps and then u see a Z, and that process is a zombie?

thanks

Right, state Z means zombie.

FWIW: On many UNIX and Linux systems zombies will not go away with a kill signal (or even kill -9) and the only way to clear them off the process table is to reboot.

Thanks for the help on how to find a zombie program. But what I really need is to find out if/when a program is no longer responding to the system so then it can be reset automatically.

Is there a way to find out about a none responding program?